Society for Neuroscience Washington, D.C.
Nov. 12-16, 2005
R.R. Ramirez; D.L. Weber; A. Delorme; C.L. Dale; G.V. Simpson; S. Makeig , Swartz Center for Computational Neuroscience, Institute for Neural Computation University of California San Diego, La Jolla CA, and Department of Radiology, UCSF, San Francisco, CA
Cortical networks involved in directed spatial attention and target detection: recursive adaptive minimum-norm ICA source imaging
Advanced magnetoencephalography (MEG) source analysis techniques were used to characterize the spatiotemporal dynamics of neural networks involved in cued visual-spatial attention and target detection. Magnetic fields were measured with a 275-sensor array (CTF) while human subjects performed a cued visual-spatial attention and target detection task. An arrow directed covert attention to the right or left lower visual field, where a grayscale sinusoid grating stimulus appeared after 1 sec with equal probability at the cued or non-cued location. Gratings were either oriented 5 deg (non-target) or 20 deg(target) from vertical. Subjects pressed a button in response to targets, only if they appeared in the cued location. The MEG time series were decomposed into temporally independent components and their associated spatially stationary, non-orthogonal sensor maps by infomax ICA in EEGLAB. The current density distribution that generated each component map was estimated using a noise-regularized recursive adaptive minimum-norm inverse algorithm applied to the ICA sensor maps (RAM-ICA) and constrained to the subsampled cortical surface (Freesurfer, Brainstorm). The event-related spectral perturbation (ERSP) and inter-trial coherence (ITC) of each independent activation function were computed for each condition using a complex Morlet wavelet transform. Confidence intervals, relative to pre-stimulus baseline and between conditions, were computed by bootstrap. Results indicate that directed attention involves contralateral alpha desynchronization and ipsilateral alpha synchronization in certain occipital sources. A source localized to the left motor cortex representation of the right hand was activated after target detection. Activity of a distributed source, localized to the left MFG, STS, IPS, and SMG was modulated by cues to attend either hemifield and by subsequent test stimuli.